Very Computer

1) To me, and to many PC, Mac, NeXT, and even Amiga users, True Colour is
   associated with 24bit graphics, that is 16777216 colours.  The
   Falcons palette is 18 bit (262144 colours).  Not to detract from the
   Falcon (which is after all a very nice machine) this is but a
   fraction of the number of colours required for True Colour.

2) What I suspect Atari mean by "True colour" is that every pixel on the
   screen is capable of being defined independantly of all other pixels.  
   Even this though is not quite true.  Thus far all reports of the Falcon
   have indicated 65536 colours as the limit for the "True colour" modes.  
   For a low-res (320x200) screen this holds true - there are only 64000
   pixels on such a screen.  However, for any mode above this - 320x480  
   for example - there considerably more than 65536 pixels on screen,
   yet still Atari claim true colour.

I realise that Atari want to sell as many Falcons as possible, but they
seem to have gone overboard with hype, giving the consumer an inaccurate
view of a machine which, from all reports, ought to be able to sell on
it's own virtues.  As an example, many people seem to believe, from the
material supplied by Atari, that the Falcon has an internal genlock, when in
terms of genlocking ability it has little more than the Amiga (such as
external synching) in that area.  Whilst I hope the Falcon sells, I do
not ultimately believe that Atari are being entirely fair to the consumers
who will in the end purchase the machines.

        Mark

--
? As long as the youngest child is free to dream ? Mark Steyn :          ?


?              Knowledge is no virtue and innocence no vice              ?

Hmmm. I'm not so sure that this isn't just nitpicking. To be able to display 65k
colours on screen is more than enough for most people. Sure, running at 320x480
will give you more pixels (over 2x) than colours, but be honest about this, on a
153k pixel screen, aren't you going to want a particular colour to be
displayed more than once ? I wish our workstations had that sort of capability.

What true-colour means to me (and I expect to most people) is that I can have
lots and lots of colours on screen. Not that I can have 16 Million (24 bit), or
indeed 65k (16 bit), just lots. I'm not sure I can think of that many
applications that use 65k colours let alone 2^24.

As an aside, I have a gif picture (Lush.gif), which is a photo-realistic picture
of lots of lush vegetation. It takes 253 colours out of my 256 on the Workstation.
When I quantise that down to 64, the effect is still stunning. In fact taking it
down to 32 colours makes it merely fantastic. Obviously it does look better in
253 colours, but I think the law of diminishing returns starts to apply. You need
a disproportionate amount of colours to make a difference over 65k. What I'm
trying to say is that I'm not sure anyone can tell the difference between a 24
bit image and a 16 bit image.

Simon.

|>   That's missing the point, Simon.  I do think Atari is being misleading
|> on this one.  The Falcon has VGA graphics.  That's all.  That's not bad,
|> it's what's made the PC the #1 game machine in the world, but it's not
|> True Color.  True Color means that you have at least 8 bits each for red,
|> green, and blue (heck, the VDI defines 10 bits each, even tho' no color
|> board exists that goes that high!), with each pixel on the screen being
|> able to take on any color at any time.
|>  
|>   Color registers are useful, but a fast color-specific bitblit can do
|> the same effects.  It's not whether or not your app uses 64K colors or
|> 16.7M colors.  It's the ability to select from that massive palette for
|> the precise color you need (and when doing rendering and digitizing,
|> 24-bit is just barely sufficient sometimes.  30-bit would have been
|> better. (and we could have had 2 bits for alpha-channel then)).

Fair enough. I guess we have different requirements. I must admit that I'd not
thought of raytracing when I wrote what I did. I guess you can raytrace that many
colours. I was more interested in what I would use it for, day to day, (which,
you'll be surprised to learn, doesn't include raytracing :-)

|>   I do intend to buy an F030, but I'm doing it with the understanding
|> that it's VGA.  That's fine with me for now, but eventually I'll be
|> giving it a REAL color board.  The sound, the OS, and the SCSI-2 are my
|> main reasons for buying it.

It is slightly better than VGA isn't it ? I didn't think VGA could put 65k
colours on the screen at a time ? I agree that if you define true colour as
8-8-8 R-G-B then the F030 isn't adequate. The thing is that it will be enough
for most people, which is *my* point. We're talking (if you believe Atari)
about the bottom-of-the-range machine here. Presumably (a BIG presumption) the
040/060? will have better graphics modes.

I agree that the system seems to have a lot going for it. In fact I consider
this to be the reason why the cacophony that normally surrounds Atari machines
has been directed solely at the graphics. Most Atari computers take a *lot* more
flak than the Falcon has. It's the first interesting machine for a long time...

I would add the DSP chip to the reasons for buying it, though, I've been
reading about them, and talking to people who use them, and they're wicked
chips. If I buy an F030, it'll probably end up being a DSP-based computer that (occaisonally :-) lets the 030 have a lookin :-)

Simon.

|> "Hey, does anyone with an advance model know how hard it is to install in
|> an AT-style case?  And if DEKA or Z-Keys works with it, so I can hook up
|> my real keyboard?"

  Color registers are useful, but a fast color-specific bitblit can do
the same effects.  It's not whether or not your app uses 64K colors or
16.7M colors.  It's the ability to select from that massive palette for
the precise color you need (and when doing rendering and digitizing,
24-bit is just barely sufficient sometimes.  30-bit would have been
better. (and we could have had 2 bits for alpha-channel then)).

  I do intend to buy an F030, but I'm doing it with the understanding
that it's VGA.  That's fine with me for now, but eventually I'll be
giving it a REAL color board.  The sound, the OS, and the SCSI-2 are my
main reasons for buying it.

"Hey, does anyone with an advance model know how hard it is to install in
an AT-style case?  And if DEKA or Z-Keys works with it, so I can hook up
my real keyboard?"

OK for the 1st point, but the second is just 'brain-damaged' !

        Stefan

BTW, how about this Falcon newsgroup that was proposed some weeks ago?
Still a good idea IMO - these threads start getting a bit long-winding
and nit-picking (OK, I admit, in this case I contribute myself, so I'm
guilty... no flames :)

--

Dept. of Philosophy                      *  in the nightmare, in the flood
University of Utrecht                    *  But I try.
The Netherlands                          *

--

University of New Hampshire Research Computing Center         -><-
Prefector of the order of St Gulik, Legion of Dynamic Discord -><-

 True colour was always a subjective term as far as I knew. It meant
having sufficient numbers of different colours on the screen that the eye
couldn't pick them out seperately.

  If there aren't 16 million pixils on the screen then 16 million colours
is useless.  The human eye can't tell even a fraction of the reds apart
from each other.

The falcon has the necessary number of colours needed to fool the eye
into believing what's on the screen is a life-like or photo-quality
image. That's true colour.  More colours than that is just
techno-overkill.  

It's like saying your machine is a gadzillion Mhz and better for it, when
all it can do is wordprocess at 60 words a minute.  

If you don't need it or can't use it, it's just extra junk.
Extra junk you end up paying big bucks for.

Quote:> The colour of a pixel is independant of the colours of all other pixels
> if and only if there are more colours available than there are pixels.

-----------------------------------------------------------------------------
"Now You're Playing With Power Without The Price!"  Glen Sescila   |||
                                                                  / | \
My opinions do not necessarily reflect those of my employer and probably are

-----------------------------------------------------------------------------

In the VGA mode, each pixel is represented by 8-bits.  These 8-bits
are an index to a CLUT.  The entries in the CLUT are 18-bits long.  To
determine the color of a pixel, you first must use the 8-bit value as
an index to a table (CLUT).  The color of that pixel is the
corresponding value in the table.

Hence, you can specify a total of 256 (2^8) colors out of 262,144
(2^18).  In other words, the 8-bits of a pixel _indirectly_ specify
the color of the pixel.

By contrast, the True Color Mode does not use a CLUT.  Instead, each
pixel is represented by 16-bits which _directly_ specify the color of
that pixel.  In other words, instead of being an index to a table, the
bits represent the _true color_ of that pixel.  With 16-bits, you can
have a total of 65,536 (2^16) colors.

That's why Atari uses the term True Color.  The bits of a pixel
represent the true color of that pixel instead of being an index to a
color table.

I personally like this definition.  It's very logical and consistent.

The Falcon can therefore be said to have 16-bit True Color as opposed
to 24-bit True Color.  By the way, I think some NeXT workstations have
a 12-bit True Color mode (and they actually used the term True Color
to describe this mode in their literature).

--Gerry

Now the size of the screen does not matter most workstation screens
are 1024 * 1024 or  bigger if the hardware supports  true color  the
max number of pixels on screen is only 1048576. To have all 16.7 mil
colors on the screen at once would need a screen resolution of 4096 *
4096. For most purposes today the use of psuedo color systems ie
a limited number of colors from 16.7 million colors is quite adiqute
as in most cases now the limited number of colors is still larger than
the display can handle.

ie 320x200 colors can be true color if the under lying palett is 16.7
million colors.

TRUE COLOR := Could you have 16.7 million colors on the screen were
              the screen resolution large enough.

              YES = True Color.
              NO  = Psuedo Color if limited color range ie 256,000 colors max
                    from 16.7 palette.

Gary.

Quote:>The falcon has the necessary number of colours needed to fool the eye
>into believing what's on the screen is a life-like or photo-quality
>image. That's true colour.  More colours than that is just
>techno-overkill.  

--
Annius Groenink  --  Universiteit Utrecht  --  PTT Research Leidschendam

                                         10 GOSUB 10000: REM  DISCLAIMER

Quote:> When there are more pixels than available colours, it's not true that
> the colour of each pixel is independant of the colours of all other
> pixels. And I think it's obvius that this is what the original poster
> intended to say (and _did_ say in fact, when one reads the first quoted
> line).
> The colour of a pixel is independant of the colours of all other pixels
> if and only if there are more colours available than there are pixels.

You only need more colours than pixels if you want the colour of each
pixel to be "different from" that of every other pixel.

If the colour of a pixel is "independent of" all the other pixels, it
just means that you can change the colour of a pixel without changing
any other pixels at the same time.  It doesn't matter if two pixels are
the same colour, so long as you can change them both separately.  You
don't need more colours than pixels to achieve this.

Or to put it another way, if you are digitizing a picture, True Colour
means that for each pixel you can select the nearest colour out of your
32768, 65535, 16M or whatever, without having to consider what colours
you have/will set other pixels to (i.e.  independently), and without
being concerned with the management of a Colour Look-Up Table or
Pallette.

Quote:> BTW, how about this Falcon newsgroup that was proposed some weeks ago?
> Still a good idea IMO - these threads start getting a bit long-winding
> and nit-picking (OK, I admit, in this case I contribute myself, so I'm
> guilty... no flames :)

Tony.
--
Tony Mountifield     (G4CJO)       | Microware Systems (UK) Ltd.

(or:  ...!uknet!mwuk!tony)         | Tel: 0703 601990   Fax: 0703 601991
------------------------------------------------------------------------
** Any opinions are mine, not Microware's - but you knew that anyway. **
------------------------------------------------------------------------

All that True Colour means is that there is no colour lookup table. The
Sinclair QL had true colour, ok, so it was 4 bit true colour.. so what!

Quote:

>Now the size of the screen does not matter most workstation screens
>are 1024 * 1024 or  bigger if the hardware supports  true color  the
>max number of pixels on screen is only 1048576. To have all 16.7 mil
>colors on the screen at once would need a screen resolution of 4096 *
>4096. For most purposes today the use of psuedo color systems ie
>a limited number of colors from 16.7 million colors is quite adiqute
>as in most cases now the limited number of colors is still larger than
>the display can handle.

>ie 320x200 colors can be true color if the under lying palett is 16.7
>million colors.

>TRUE COLOR := Could you have 16.7 million colors on the screen were
>              the screen resolution large enough.

>              YES = True Color.
>              NO  = Psuedo Color if limited color range ie 256,000 colors max
>                    from 16.7 palette.

>Gary.

Now, on the point of "is 16 bit colour any good?", yes it is.. it is close
to the colour resolution of the human eye. The only problem I see is with
grey-scale images, but two bits of brightness resolution should not make
much difference at all.

I'm sorry if this seems like a flame.. it isn't really.. I'm just a bit fed
up with the way this discussion is just bouncing between a couple of people
and getting nowhere!

Steve

--
Addresses:-

Just wondering,

Sander SToks